On Mon, Jun 15, 2026, at 16:37, 
[email protected] wrote:
> Bron,
> 
> Thanks for confirming the originator framing.
> 
> On the nonce: the SES message ID is a natural fit. It is already opaque to 
> outsiders, it is the S3 object key, and it is a direct lookup into storage we 
> already have -- no extra bookkeeping. One thing worth noting: a nonce past 
> the 30-day retention window is a transparency statement, not a gap. The S3 
> object is gone by design at that point. That is a privacy feature.
> 
> On Message-ID: yes, we do rewrite it. We also scrub References and 
> In-Reply-To, rewriting them to our own domain namespace. Thread identifiers 
> in the originating provider's format would otherwise propagate sender and 
> provider information across the relay boundary.
> 
> On the considerations text: intentional re-origination by a 
> privacy-preserving intermediary should be acknowledged as a legitimate 
> deployment pattern, so that an assumed missing chain is not read as evasion 
> by downstream deployment or reputation guidance. There are many people who 
> want this level of privacy in their email. Recipient privacy is a first-class 
> use case. Happy to propose text if that would help move it forward. 

I really don't think there's any text required at all, but if there was it 
would belong in draft-herr-dkim2-bcp.

You're basically creating a brand new message out of some cherry-picked parts 
of a message which came into your system.  It's not "the same email" in any 
meaningful sense, and you're deliberately NOT providing a chain of custody 
through your systems - the opposite of what DKIM2 (or indeed ARC, or even just 
the Received headers) is designed to do.

So you're doing something completely different to DKIM2; up until you send the 
message out into the rest of the world again, in which case it's a new message 
and you quite rightly take ownership of it.

Bron.

--
  Bron Gondwana, CEO, Fastmail Pty Ltd / Fastmail US LLC
  [email protected]

_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to